Peak + Accumulation: A Proxy-Level Scoring Formula for Multi-Turn LLM Attack Detection

ArXi:2602.11247v2 Announce Type: replace-cross Multi-turn prompt injection attacks distribute malicious intent across multiple conversation turns, exploiting the assumption that each turn is evaluated independently. While single-turn detection has been extensively studied, no published formula exists for aggregating per-turn pattern scores into a conversation-level risk score at the proxy layer -- without invoking an